Built a multi-agent AI mental health assistant (7 agents, backend automated, no-code stack)

Been working on this little side project and finally got it to a working version. It’s an AI-powered mental health assistant — not just a chatbot, but a system that can retrieve user history, analyze input, access data in real-time, and suggest personalized treatment plans.

UI

Chat

Tech stack: Loveable + Momen

How it’s structured:

• It uses 7 specialized AI agents, each responsible for a niche task — chat, generate professional responses, summarize user info, classify intent, etc.

Agent Team

• The main agent (the chat one) will call other agents in the backend via automated workflows.
• It keeps track of user data (symptoms, conversations, medical history) and updates it in real time — all triggered automatically.
• Everything runs in the backend to reduce manual steps and minimize errors.

How it’s built:

• Started by drafting the UI with Loveable AI — it auto-generated a 7-page interface from a product brief, which saved me time.

(Didn’t use it for the live app though — good for prototyping, but I wanted more control for complex backend workflows.)

• Rebuilt the UI and database in Momen, since I needed deeper control over data flow and backend logic.
• The entire AI agent system and backend workflows were built in Momen as well. So I can make the agents collaborate with each other. The main chat agent invokes backend workflows to call other agents when needed.

Entire flow looks like this: the user sends a message, the system: → pulls in the latest user data→ triggers the right agent(s) based on the input→ responds in real-time→ quietly summarizes and updates everything in the background.

FlowChart

It’s still an MVP, but the multi-agent setup + automated backend feels pretty scalable.This was a super fun build and I learned a lot about orchestrating AI workflows.

Would love any feedback or thoughts on how to improve this.